Alveolar macrophages play a crucial role in initiating the inflammatory response in allergic asthma through the cross-linking of the low affinity IgE receptors (FcεRIIb or CD23) by IgE-allergen immunocomplexes. We have previously shown that CD23 cross-linking in monocytes and U937 cells targets IκBα, leading to the activation of the transcription factor NF-κB. We demonstrate in this paper that CD23-initiated signaling in U937 cells leads to hyperphosphorylation of IκBα at Ser32/Ser36 residues. Overexpression of a dominant-negative IκBα transgene containing mutations at Ser32/Ser36 completely inhibits degradation of IκBα, NF-κB activation, and gene transcription that follows CD23 cross-linking. Investigation of the second messengers mediating the CD23-dependent activation of NFκB demonstrates that IκB kinases (IKKs) but not p90rsk are selectively activated following CD23 cross-linking and mediates the phosphorylation of IκBα. Cotransfection experiments with an IKKβ negative dominant completely inhibit CD23 induced NFκB activation. Furthermore, the activation of tyrosine kinase(s) by CD23 is required for the induction of IKK activity, IκBα degradation, and NF-κB nuclear translocation. Taken together, our results show that CD23 cross-linking in the monocytic lineage induces tyrosine kinase activation followed by activation of IKK, which phosphorylates IκBα at the N-terminal domain (Ser32/Ser36), inducing its degradation, NF-κB activation and gene transcription.
Alveolar macrophages (AM) are essential for the initiation and perpetuation of the inflammatory reaction that characterizes bronchial asthma. In atopic individuals, AM express high levels of the surface low affinity IgE receptor FcεRIIb or CD23 (1). CD23 on monocytes is a single chain receptor with a small intracytoplasmic domain that differs from that in B cells (2). Activation of AM by binding of IgE-allergen immunocomplexes to CD23 induces the secretion of arachidonic acid metabolites, oxygen radicals, and multiple monokines that are responsible for the damage of the bronchial epithelium and for the activation and recruitment of other immune cells (3). The mechanism by which CD23 transmits intracellular signals that lead to monocyte/macrophage activation is not known and the molecules that associate to its intracytoplasmic domain in monocytes have not been identified yet. We have recently shown that cross-linking of CD23 in both human monocytes and the promonocytic cell line U937 leads to nuclear translocation of the transcription factor NF-κB, which directly correlates with increased transcription of the inflammatory cytokines IL-1β and TNF-α in human monocytes (4).
NF-κB is a ubiquitously expressed transcription factor composed of homo- or heterodimers of DNA-binding proteins from the rel family, which control transcription of genes encoding many inflammatory cytokines. NF-κB is present in an inactive form in the cytoplasm through its interaction with IκB. Phosphorylation of IκB triggered by activation signals is necessary for its ubiquitination and degradation, allowing for the release of NF-κB (5). Several members of the IκB family of NF-κB inhibitory proteins have been described (IκBα, IκBβ, IκBγ, IκBε, p105, p100, bcl-3) (5). IκBα contains several residues that can be potential targets for inducible phosphorylation. Serines in the amino terminus of the molecule (Ser32 and Ser36) are phosphorylated by cytokine-activated kinases present in the IκB kinase (IKK)3 signalsome (6, 7, 8) or by the mitogen-activated p90rsk kinase (9, 10). Recent reports indicate that IκBα can also be inducibly phosphorylated at Tyr42, by means of a yet unidentified tyrosine kinase (TK) (11, 12). The IκBα/NF-κB equilibrium is regulated not only by kinases that inducibly phosphorylate IκBα, but also by phosphatases that inactivate IKK, which result in the inhibition of IκBα phosphorylation and degradation. This finding is supported by the effect of okadaic acid (OA), an inhibitor of protein-phosphatase 2A (PP2A), which induces IκBα N-terminal phosphorylation and subsequent NF-κB activation (13).
In a previous report it was demonstrated that CD23 cross-linking led to NF-κB activation as a result of IκBα hyperphosphorylation, which was dependent on TK activity (4). The exact IκBα residues hyperphosphorylated during CD23 signaling (e.g., Tyr42, Ser32, Ser36, or others) remained unknown. Their determination would then provide a clue for the identification of the kinase(s) responsible for IκBα hyperphosphorylation. In the present study we have demonstrated that Ser32 and Ser36, but not Tyr42, are the residues targeted for IκBα hyperphosphorylation following CD23 cross-linking. Of the kinases known to target IκBα Ser32/Ser36, IKK, but not p90rsk, is responsible for IκBα hyperphosphorylation during CD23 signaling. Overexpression of an IKKβ negative dominant completely blocks CD23-induced NF-κB activation, suggesting an essential role for this kinase in the signal transduction of CD23 in monocytes. Moreover, we show that a TK, upstream of IKK, is required for the CD23-mediated IκBα degradation and hence, NF-κB activation.
Materials and Methods
Cell lines, induction, and analysis of CD23 expression
The promonocytic cell line U937 was obtained from the American Type Tissue Collection (Manassas, VA) and cultured in RPMI medium supplemented with 5% heat-inactivated FBS (Invitrogen, Purchase, NY), 100 U/ml penicillin/streptomycin, and 2 mM l-glutamine. Monocytes were purified from human buffy coats by Ficoll-Hypaque gradient separation and then adherence to plastic tissue culture flasks overnight in RPMI medium with 10% human AB serum (BioWhittaker, Walkersville, MD). Nonadherent cells were removed and the adherent population was incubated in the same medium for additional 48 h. U937 cells grown at a density of 0.25 × 106 cells/ml, or monocytes adhered to tissue culture flasks, were incubated with 10 ng/ml of recombinant human IL-4 (R&D Systems, Minneapolis, MN) for 48 h at 37°C. Surface CD23 expression was analyzed in a flow cytometer (FACScan, Becton Dickinson, Franklin Lakes, NJ) using PE-conjugated anti-CD23 or the PE-isotype matched control Ab (Becton Dickinson) by standard procedures.
Cells were incubated at 5 × 106 cells/ml in 6- well plates with 20 μg/ml affinity-purified monoclonal human IgE (Fitzgerald Industries, Concord, MA) for 1 h at 37°C. Cells were then centrifuged and resuspended in fresh culture medium containing 20 μg/ml affinity-purified goat anti-human IgE Ab (GαHIgE, Fitzgerald Industries) at 37°C for 10–30 min.
U937 cells were incubated at 5 × 106 cells/ml in 6-well plates with medium containing 10 ng/ml human TNF-α (R&D Systems) for 10 min.
A total of 4 μM of the TK inhibitor herbimycin A (HA; Calbiochem, San Diego, CA) reconstituted in DMSO was added to the cell cultures 18 h before the CD23 cross-linking; 3 μM of the protein kinase C inhibitor Bisindolylmaleimide I (GF 109203X, Calbiochem) was added to the cell cultures for 1 h before and during CD23 cross-linking; 30 μM of the MEK1 inhibitor PD 098059 (Alexis, San Diego, CA), reconstituted in DMSO was added to the cells 1 h before and at the time of CD23 cross-linking; 30 nM of the PP2A inhibitor OA, reconstituted in ethanol was incubated with the cells for 5 h prior and during the CD23 cross-linking. In the experiments were HA was used, 1 mM of the tyrosine phosphatase inhibitor NaVO4 (Sigma, St. Louis, MO) was added to the lysis buffer (buffer B).
pCMV2-Flag-IκBαwt consisted of the full-length sequence of human IκBα (12) obtained from Cetus (Emeryville, CA), cloned into the SmaI-HindIII sites of pCMV2-Flag (Kodak, New Haven, CT) to generate N-terminally Flag-tagged IκBα. Flag-IκBαwt was used as a template for point mutations using standard PCR techniques.
Flag-IκBαS was derived by overlap site-directed mutagenesis (14) with Taq DNA polymerase and consisted of the full-length sequence of IκBαwt in which Ser32 and Ser36 were substituted by alanine residues. The primers used to generate these mutations were: 5′-GACGCAGGCCTGGACGCAATG-3′ (sense) and 5′-CATTGCGTCCAGGCCTGCGTC-3′ (antisense).
Flag-IκBαT was generated by site-directed mutagenesis as above and consisted of the full-length sequence of IκBαwt in which Tyr42 was substituted by phenylalanine. The primers used to generate this mutation were: 5′-GAGGAGGCCGAGCAGATGGTC-3′ (sense) and 5′-GACCATCTGCTCGGCCTCCTC-3′ (antisense).
Flag-IκBαwt, Flag-IκBαS, and Flag-IκBαT were subcloned into the EcoRI site of pSFFV-Neo (15). Sequence orientation and the presence of mutations were verified by DNA sequencing.
We have consistently observed a faster migration of Flag-IκBαT than other Flag-IκBα constructs in Western blot analysis. We have confirmed that the length of all the Flag-IκBα inserts is identical by electrophoresis of the DNA following EcoRI restriction and by sequencing (data not shown). Therefore, we conclude that the faster migration of Flag-IκBαT in Western blots is due to a conformational change of the protein induced by the substitution of tyrosine by phenylalanine.
Plasmid κB-luc contains three tandem copies of the HIV long terminal repeat κB motif, cloned upstream of the minimal conalbumin-luciferase (con-luc) promoter reporter gene. The pRL-TK vector contains the herpes simplex virus thymidine kinase promoter region upstream of the Renilla-luciferase reporter gene (Promega, Madison, WI). IKKβ kinase dead (IKKβKD) was kindly provided by J. Woronicz (Tularik, San Jose, CA). IKKβKD contains the IKKβ cDNA with a point mutation in Lys44 to alanine, cloned in the pRK5 expression vector, and has been previously shown to function as a dominant negative (14). The expression vector pCMV2 was obtained from Kodak.
Ten million U937 cells, grown at 0.5 × 106 cells/ml, were resuspended in 250 μl RPMI containing 5% FBS. Plasmid DNA in 50 μg of pSFFV-Neo, linearized by SalI, was added to the cells, which were then electroporated at 350 V for 10 ms. Transfected cells were resuspended in selection media containing G418 (Calbiochem) at 1 mg/ml, 24 h after the electroporation. Cells were cloned by limiting dilution about 3 wk after the transfection. Stable integration and expression of the transfected gene within each monoclonal population was assessed by Western blot analysis of whole cell extracts and immunoblotting with anti-Flag Abs.
U937 cells were treated with 10 ng/ml of IL-4 for 24 h. Cells (2.5 × 106) were transfected with 0.5 μg of the kB-luc reporter plasmid, 0.5 μg of the pRL-TK vector, and 1 μg of either the IKKβKD or a CMV2 plasmid, using the Vectamidine method, by following the manufacturer’s recommendations (Accurate Chemical & Scientific, Westbury, NY). Twenty four hours after transfection, CD23 was cross-linked with IgE followed by GαHIgE, as described above. Luciferase levels were measured with the Promega dual-luciferase reporter assay system 4 h after CD23 cross-linking was performed. Transfections were normalized with the luciferase activity of the pRL-TK vector, following the manufacturer’s protocol (Promega).
Nuclear extraction and gel mobility shift assay
Nuclear protein was extracted by using a modification of the method of Dignam et al. (16). Briefly, 107 cells were washed with buffer A (10 mM HEPES, 1.5 mM MgCl2, and 10 mM KCl). Cells were then lysed with buffer B (buffer A containing 0.1% Nonidet P-40) for 4 min at 0°C and washed with buffer A. The nuclear pellet was resuspended in 20 μl of buffer C (20 mM HEPES, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl2, and 0.2 mM EDTA) and incubated at 4°C for 30 min in a rotator. Following centrifugation, the supernatant was diluted with 40 μl of buffer D (20 mM HEPES, 20% glycerol, 0.05 M KCl, and 0.2 mM EDTA) and stored at −70°C. All buffers contained 0.5 mM PMSF, and buffers B, C, and D also contained 0.5 mM DTT, 2 μg/ml aprotinin, 2 μg/ml leupeptin, and 2 μg/ml pepstatin. Proteins were quantified by using the Bio-Rad (Richmond, CA) protein assay. For mobility shift assays, 3 μg of protein extract was incubated with [γ-32P]ATP-labeled double-stranded oligodeoxynucleotide probe at room temperature for 15 min with 3 μl of binding buffer and 1 μg of poly(dIdC) (Pharmacia, Piscataway, NJ). The binding buffer contained 100 mM HEPES, 300 mM KCl, 20% Ficoll, 0.05% Nonidet P-40, and 0.5 mg/ml BSA. The binding reaction was analyzed by electrophoresis in a nondenaturing 5% polyacrylamide gel and visualized by autoradiography. The oligonucleotide used in the binding reaction corresponded to the NF-κB binding sequence present within the enhancer of the HIV long terminal repeat (5′-ACAAGGGACTTTCCGCTGGGGACTTTCCAGGGA-3′). The double-stranded probe was end-labeled with [γ-32P]dATP using polynucleotide kinase.
Cytosolic extracts and Western blots
Cytosolic proteins were obtained from the supernatant of the cells lysed with buffer B (see above). Crude extract cytosolic proteins (20 μg) were denatured by boiling, separated by 10% SDS-PAGE, and transferred to Immobilon-P membranes (Millipore, Bedford, MA) by standard procedures. Immunoblotting was done with polyclonal rabbit anti-human IκBα Ab (prepared by immunization with recombinant GST-IκBα protein, described below), anti-IKKα (Santa Cruz Biotechnology, Santa Cruz, CA), or anti-p90rsk Ab (Santa Cruz Biotechnology). To characterize the level of expression of Flag-IκBα in transfected cells, an anti-Flag mAb (IBI, New Haven, CT) followed by incubation with rabbit anti-mouse IgG (Pierce, Rockford, IL) and anti-HRP (Amersham, Arlington Heights, IL) were used for immunoblotting. Reactive proteins were visualized by using an enhanced chemiluminescence (ECL) Western blotting detection kit (Amersham). To ensure equal amounts of protein loading, Western blot membranes were also blotted with an affinity-purified polyclonal rabbit anti-human β-actin Ab (Sigma).
RNA protection assay (RPA)
Total RNA was obtained from 5 × 106 U937 cells by the RNAzol method, following the manufacturer’s recommendations (Tel-Test, Friendswood, TX). Six micrograms of RNA from the U937 clones transfected with the Flag-IκBαwt and Flag-IκBαS constructs was subjected to RPA with the Riboquant method using the hCK-3 multiprobe template set, following the manufacturer’s recommendations (PharMingen, San Diego, CA).
Immunocomplex kinase assay
Cytosolic proteins (100 μg) from whole cell extracts were immunoprecipitated by incubation with anti-IKKα or anti-p90rsk Ab for 1 h, followed by incubation with Protein A agarose beads for an additional hour (Life Technologies, Gaithersburg, MD) at 4°C, with slow rotation. The immunoprecipitations were performed in 0.5 M NaCl buffer. The beads were washed three times with 0.5 M NaCl buffer followed by one wash with kinase assay washing buffer (50 mM Tris-HCl (pH 7.4) and 40 mM NaCl). The beads were mixed with 15 μl of kinase buffer (20 mM HEPES (pH 7.4), 2 mM magnesium chloride, 2 mM manganese chloride, 10 μM adenosine triphosphate, 10 mM NaF, 10 mM p-nitrophenylphosphate (PNPP), 10 mM β-glycerophosphate, 300 μM NaVO4, 2 μM PMSF, 10 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 mM DDT). The in vitro kinase assay was performed by incubating the beads in the kinase buffer containing 2 μg of GST-IκBα1–53(1–53) fusion protein substrate and 1 mCi of [γ-32P] ATP. GST-IκBα1–53(1–53) fusion protein was obtained by amplifying the amino-terminal sequence of IκBα1–53(1–53) using primer A (CGGGATCCATGTTCCAGGCGGCCGAG), as the 5′ sense primer, creating a BamHI site upstream of the coding sequence, and primer B (GGAATTCCTCAGCGGATCTCCTGCAGCT) as antisense primer, creating an EcoRI site downstream of the coding sequence. Following digestion with BamHI and EcoRI, these sequences were ligated into pGEX-KG (derived from pGEX-2T from Pharmacia Biotech, Piscataway, NJ). These constructs were transformed into Escherichia coli DH5α cells, which were grown exponentially, and after 60 min of stimulation with isopropylthiogalactopyranoside (IPTG) (Sigma) cells were lysed. Proteins were isolated by affinity chromatography on gluthathione-bonded 4% cross-linked agarose (Sigma). The purity of GST-IκBα1–53(1–53) was analyzed with 10% SDS-PAGE and subsequent Coomassie blue staining. The purity of the protein was >90%. The kinase reaction was performed at 30°C for 30 min and stopped by adding 4× SDS-PAGE sample buffer. The proteins were separated by SDS-PAGE and transferred to Immobilon-P membrane. The top part of the membrane was used for immunoblots of IKKα or p90rsk; on the bottom part of the membrane, the amount of GST-IκBα and the levels of its phosphorylation were visualized by staining with Coomassie blue and autoradiography, respectively. P90rsk activity was quantitated by AMBIS Radioanalytical Imaging System densitometry (AMBIS, San Diego, CA).
CD23-induced hyperphosphorylation of IκBα occurs at Ser32/Ser36
IκBα hyperphosphorylation following cell activation has been reported to occur either at Ser32/Ser36 or at Tyr42 residues present in the N-terminal domain of the molecule. To characterize the kinases and second messengers involved in IκBα phosphorylation, we first analyzed what IκBα amino acids are being phosphorylated following CD23 cross-linking. Clones of U937 cells that stably express Flag-tagged IκBα, either containing or not containing mutations at the different phosphorylation sites (Ser32/Ser36 and Tyr42) were used for this purpose. We first verified that the Flag-tagged IκBα expressing U937 clones behaved like the parental cells with respect to the CD23 expression. U937 clones were treated or not with IL-4 to increase surface CD23 expression, which was analyzed by FACS. Although there was some variability in CD23 expression depending on the cell cycle, the degree of CD23 inducibility by IL-4 was similar in the clones and the parental nontransfected cell lines (Table I).
|Cells .||Ab .||IL-4 .||% Positive Cells .|
|Cells .||Ab .||IL-4 .||% Positive Cells .|
We next analyzed whether the Flag-tagged IκBαwt expressed in the U937 clones behaved similarly to the endogenous IκBα. We have previously shown that CD23 cross-linking results in hyperphosphorylation and degradation of IκBα (4). As shown in Fig. 1, CD23 cross-linking induces the degradation of the endogenous IκBα (lanes 2, 4, 6, and 8) as well as the Flag-IκBαwt (lane 4). Having demonstrated that the IκBαwt transgene behaves in a similar manner as the endogenous IκBα, we next investigated whether mutations at Ser32/Ser36 or Tyr42 interfere with the CD23-mediated IκBα degradation. As shown in Fig. 1, Flag-IκBαT (containing a mutation in Tyr42) was degraded in a similar manner as Flag-IκBαwt and endogenous IκBα (lane 8). However, Flag-IκBαS (containing mutations at Ser32/Ser36) was not degraded following CD23 cross-linking (lane 6). These results indicate that the hyperphosphorylation of IκBα induced by CD23 cross-linking in U937 cells targets Ser32/Ser36 residues.
To correlate IκBα degradation with NF-κB nuclear translocation, gel shift assay analysis of nuclear proteins from IL-4-treated U937 cells expressing the different Flag-IκBα constructs was performed following CD23 activation. As shown in Fig. 2, NF-κB (p65/p50) was translocated to the nucleus following CD23 cross-linking in U937 clones expressing Flag-IκBαwt (lane 4), Flag-IκBαT (lane 8), or in the nontransfected parental control cells (lane 2). However, NF-κB was not translocated to the nuclei following CD23 activation in U937 clones expressing Flag-IκBαS (lane 6). The lower m.w. band and smear in the autoradiograph probably correspond to p50/p50 homodimers and degradation products, respectively.
To assess whether phosphorylation of IκBα at Ser32/Ser36 was important not only for NF-κB activation but also for CD23-induced gene transcription, CD23 triggered TNF-α transcription was analyzed in the U937 clones expressing the Flag-IκBα constructs by RPA. As shown in Fig. 3, CD23 cross-linking induced TNF-α transcription in both clones overexpressing Flag-IκBαwt and Flag-IκBαS (lanes 3 and 6, respectively). However, the amount of TNF-α mRNA was lower in the Flag-IκBαS clone (compare lanes 3 and 6). Equal amounts of total RNA were analyzed in the RPA, as assessed by protection of the housekeeping gene L32 (lower panel). The small induction of TNF-α transcription by CD23 in the clone expressing Flag-IκBαS can be explained by CD23-induced phosphorylation of endogenous IκBα in these cells (see Fig. 1) or by the activation of another NF-κB-independent pathway.
Overall, these data confirm that CD23-induced hyperphosphorylation of IκBα at Ser32/Ser36 is essential for NF-κB activation and is important for TNF-α transcription in U937 cells.
IKK mediates IκBα hyperphosphorylation following CD23 cross-linking
To identify the kinase(s) responsible for IκBα phosphorylation following CD23 cross-linking, we have focused in the kinases that are known to phosphorylate IκBα at Ser32/Ser36 residues, such as IKK (6, 7, 8) and p90rsk (9, 10). To study the role of IKK, its kinase activity was measured in an in vitro kinase assay in which immunoprecipitated IKK was incubated with GST-IκBα1–53(1–53) as a substrate. As shown in Fig. 4 A, the basal level of IKK activity (GST-IκBα-P) in U937 cells (lane 1) did not change following IL-4 treatment for 48 h (lane 3), but significantly increased following CD23 cross-linking (lane 4). The amount of IKK activity induced by CD23 cross-linking was similar to that induced by TNF-α stimulation (lane 2). The increased IκBα phosphorylation was not due to unequal loading of the substrate in the in vitro kinase reaction, as determined by Coomassie blue staining of the membrane [GST-IκBα (Coo)], nor to differences in the amount of IKK immunoprecipitated, as demonstrated by immunoblotting of the membrane with an anti-IKKα Ab [IKKα (IB)]. Similar results were obtained with whole cell extracts from IL-4-treated primary monocytes following CD23 cross-linking (lanes 5 and 6).
The degree of in vitro kinase activity directly correlated with the degree of in vivo IκBα degradation in the same samples, as demonstrated by immunoblotting the cytosolic extracts from which IKK activity was measured with an anti-IκBα Ab [IκBα (IB), Fig. 4 B]. To ensure equal protein loading of this assay, the membrane used for IκBα (IB) was subsequently immunoblotted with an anti-β-actin Ab [β-actin (IB)]. These results indicate that CD23 induced IκBα hyperphosphorylation is mediated by the IKK signalsome in both, U937 cells and primary monocytes.
CD23-induced IKK activity is dependent on TK phosphorylation
Previous studies from our group have demonstrated that the TK inhibitor HA completely inhibited IκBα degradation and NF-κB activation induced by CD23 cross-linking in U937 cells (4). Having shown that the IκBα hyperphosphorylation mediated by CD23 cross-linking does not occur at Tyr42 but rather at Ser32/Ser36, we questioned whether functional TK is required for the CD23-induced IKK activation. We first studied the effect of HA on CD23-induced IKK activity by treating the cell cultures with HA for 18 h before the CD23 cross-linking. As shown in Fig. 5 A, CD23-induced IKK activity (GST-IκBα-P, lane 3) was completely blocked in the presence of HA (lane 4), as determined in an in vitro kinase assay with IKK immunoprecipitated from U937 cytoplasmic extracts and using GST-IκBα1–53(1–53) as a substrate. This result confirms the need of functional TK for the activation of IKK in the CD23 signal transduction pathway.
Because the activity of the phosphatase PP2A is inhibited by TK (17) and PP2A dephosphorylates IKKα, causing inhibition of its kinase activity (13), we questioned whether the requirement of TK activity for the CD23-induced IKK kinase activity was regulated by PP2A. To address this question, we used OA, a known activator of NF-κB and selective inhibitor of PP2A (13). U937 cells were incubated with OA before and during CD23 cross-linking. As indicated in Fig. 5, lane 2, OA induced increased IKK activity (Fig. 5,A) and IκBα degradation (Fig. 5,B) but did not revert the effects of HA inhibiting CD23-induced IKK activity and IκBα degradation (Fig. 5, A and B, respectively, compare lanes 4 and 5). These results indicate that PP2A is required for regulating the basal but not the CD23-induced IKK activity.
To determine whether TK activity is a general requirement for IKK activation, similar experiments were performed upon stimulation of U937 cells with TNF-α. As shown in Fig. 5,A, TNF-α induced increased IKK activity (lane 7) over baseline (lane 6), which was not significantly inhibited by HA pretreatment (lane 8). These results indicate that TK activity is required for the activation of IKK in the CD23 but not the TNF-α signal transduction pathway. Equal loading of immunoprecipitated IKKα and of the substrate for the in vitro kinase reaction (Fig. 5,A), as well as equal protein loading in the IκBα immunoblot (Fig. 5,B) was demonstrated as described for Fig. 4.
IKKβ is essential for CD23-induced NF-κB activation
We have shown that kinase activity in the cell signalsome mediates CD23-induced IκBα phosphorylation. However, the entire signalsome complex, containing multiple proteins, can be immunoprecipitated with the Abs used to IKKα. Therefore, the exact nature of the kinase responsible for IκBα phosphorylation following CD23 cross-linking remained unknown. Recent reports have shown that IKKβ, but not IKKα, is essential in cytokine-mediated cell activation (18, 19). To investigate whether this is also the case in CD23-mediated NF-κB activation, IL-4-treated U937 cells were transiently cotransfected with a luciferase reporter vector driven by NF-κB binding sites (κB-luc) and either an empty (CMV2) or an IKKβ dominant negative (IKKβKD) expression vector. As shown in Fig. 6, CD23 cross-linking induced luciferase activity (10-fold activation over baseline), which was almost completely blocked by overexpressed IKKβ dominant negative. Luciferase activity was normalized by cotransfection with the pRL-TK reporter plasmid.
These results indicate that as previously shown for other receptors, IKKβ is essential for NF-κB activation in the signal transduction of CD23.
CD23 cross-linking minimally enhances p90rsk activity, which does not contribute to IκBα degradation
Recent reports have demonstrated that IκBα can be phosphorylated at Ser32 by p90rsk kinase, which lies downstream of MEK1 in the Ras-Raf-MEK-MAP (mitogen-activating protein) kinase pathway (9, 10). To investigate whether p90rsk was involved in the hyperphosphorylation and degradation of IκBα following CD23 cross-linking, its potential IκBα kinase activity was studied in an in vitro kinase assay as described in Fig. 4 for IKK. As shown in Fig. 7,A, p90rsk IκBα kinase activity (GST-IκBα-P) was induced following IL-4 treatment (lane 3) and further increased (minimally) following CD23 cross-linking (lane 6). Pretreatment of the cells with HA completely inhibited both, the basal level and the CD23-induced p90rsk activity (lane 9). As expected, TNF-α treatment did not modify p90rsk activity (lane 2), although it induced complete degradation of IκBα (Fig. 7 B, lane 2). Equal loading of immunoprecipitated p90rsk and the substrate in the in vitro kinase reaction, was determined by immunoblotting of the membrane with an anti-p90rsk Ab (p90rsk (IB)) and by Coomassie blue staining (GST-IκBα (Coo)), respectively.
To determine the relevance of CD23-induced p90rsk activity inducing IκBα phosphorylation, we performed CD23 cross-linking in the presence of PD 098059 (PD), a specific inhibitor of MEK1 (20), and GF 109203X (GF), a protein kinase C inhibitor (21). Both inhibitors decreased p90rsk activity in IL-4-treated cells to almost basal levels (lanes 4 and 5) and reduced the CD23-induced p90rsk activity to the level of IL-4-treated non-CD23-cross-linked cells (lanes 7 and 8). However, the presence of PD and GF did not affect the degradation of IκBα induced by CD23 cross-linking in vivo (Fig. 7,B, lanes 7 and 8). Equal protein loading of the membrane in Fig. 7 B was demonstrated by immunoblotting with an anti-β-Actin Ab (β-actin (IB)).
These results indicate that p90rsk is minimally activated following CD23 cross-linking, and its activity is not essential in the signal transduction of CD23 leading to NF-κB activation, since IκBα can still be phosphorylated and degraded in vivo in the presence of p90rsk inhibitors.
As summarized in Fig 8, our findings indicate that cross-linking of CD23 receptors in the promonocytic cell line U937 leads to induction of tyrosine kinase activity followed by activation of IKK in the signalsome. Activated IKK then phosphorylates IκBα at Ser32/Ser36, inducing its ubiquitination and degradation in the proteosome. NF-κB then translocates to the nucleus and activates the transcription of multiple genes. In addition, CD23 cross-linking is also followed by some increase of p90rsk activity, which by itself is not sufficient to induce IκBα degradation. The CD23-induced p90rsk activity is also dependent on a TK. PP2A seems to be involved in the dephosphorylation of IKK to down-regulate the basal activity of the signalsome, but does not play a role in the CD23 signal transduction pathway.
Other investigators have shown phosphorylation of IκBα at Tyr42 in pervanadate treated cells (11, 12). However, a specific TK responsible for this phosphorylation has not been identified. We demonstrate that even though TK activity is essential for CD23 signaling, IκBα does not get phosphorylated at Tyr42. The CD23-induced TK activity is upstream to the activation of the IKK signalsome.
This is the first report showing involvement of the IKK signalsome and demonstrating the exact sites of IκBα phosphorylation in the CD23 signal transduction pathway.
We had previously demonstrated that NF-κB was activated by CD23 signaling in monocytic cells and were unable to find concomitant activation of other transcription factors (e.g., AP-1, SP-1, and CREB) (4). This agrees with our findings that IKK, and not p90rsk, is the main kinase triggered by CD23 signaling, because p90rsk has been shown to be involved in the activation of AP-1 in vitro (22). Our results are also important in delineating a single pathway triggered by CD23 cross-linking and leading to NF-κB activation. These data are helpful to identify molecules that could be possible targets for drug development for the treatment of asthma.
We have shown that the kinase involved in CD23-induced IκBα phosphorylation is a component of the cell signalsome. This activity was studied by immunoprecipitating the signalsome with an anti-IKKα Ab. However, because the signalsome is a 700–900 kDa protein complex, it was possible that the kinase activity seen following CD23 cross-linking be due to a kinase different from IKKα, which was immunoprecipitated within this complex. Other components of the cell signalsome include IKKβ (6, 7, 8) and the recently described NF-κB essential modulator (NEMO) (23). Cotransfection studies with an IKKβ negative dominant indicate that IKKβ is essential for CD23-mediated NF-κB activation.
These results show that both CD23 and TNF-α signaling converge at the level of IKK activation. However, these two signaling pathways differ in kinetics, because TNF-α induces a faster IκBα phosphorylation than CD23 cross-linking (4). In addition, CD23 but not TNF-α pathway require upstream TK activity. Further work needs to be done to investigate other intermediate molecules such as NF-κB-inducing kinase (NIK), MEK kinases, and NEMO to elucidate the exact differences in signal transduction triggered by these cell stimuli.
We have recently demonstrated that the signal transduction of CD23 in U937 cells and human monocytes is comparable in that in both cells there is IκBα phosphorylation followed by NF-κB activation (4). In addition, we show here that CD23 cross-linking induces IKK activation in both cell types. Therefore, we can presume that the results obtained here with transfected U937 clones can be applied to monocytes and are important in understanding the pathogenesis of asthma. However, further investigation is needed to fill up the gaps in the CD23 signal transduction pathway and to determine the most relevant targets for future drugs aimed at blocking NF-κB activation, which can have an impact in controlling the inflammation in asthma.
We are grateful to all members of the C. V. Paya laboratory for helpful discussion.
This work was supported by funds from the Mayo Foundation.
Abbreviations used in this paper: IKK, IκB kinase; TK, tyrosine kinase; PP2A, protein-phosphatase 2A; GαHIgE, goat anti-human IgE Ab; MEK, mitogen-activated protein/extracellular signal-related kinase kinase; IKKβKD, IKKβ kinase dead; RPA, RNA protection assay; HA, herbimycin A; OA, okadaic acid.